System Requirements
Hardware and software requirements for running OpenClaw (formerly Moltbot) on your machine.
Cloud API only
- CPU2 cores
- RAM2 GB
- Storage10 GB
- Network1 Mbps
Suitable for: $5 VPS, Raspberry Pi 4
Small local models
- CPU4 cores
- RAM16 GB
- Storage50 GB SSD
- Network10 Mbps
Suitable for: Mac Mini M1, mid-range VPS
Large local models
- CPU8+ cores
- RAM32+ GB
- Storage100+ GB NVMe
- GPU8+ GB VRAM
Suitable for: Mac Studio, Gaming PC, Server
Native Support
- ✓ macOS 12+ (Monterey and later)
- ✓ Ubuntu 20.04+
- ✓ Debian 11+
- ✓ Fedora 36+
- ✓ Arch Linux (rolling)
Via Docker/WSL
- ✓ Windows 10 (version 2004+)
- ✓ Windows 11
- ✓ Any Linux with Docker
- ✓ Synology NAS (Docker)
- ✓ QNAP NAS (Docker)
The installer handles these automatically, but for manual setup:
Required
- • Node.js 22+ (required)
- • pnpm 8+ (package manager)
- • Git (for updates)
Optional
- • Docker (for containerized setup)
- • Ollama (for local AI models)
- • Chromium (for browser automation)
If you want to run AI locally with Ollama (free, unlimited usage):
| Model Size | RAM Required | Disk Space | Example Models |
|---|---|---|---|
| 7B | 8 GB | 4 GB | Llama 3 7B, Mistral 7B |
| 13B | 16 GB | 8 GB | Llama 3 13B, CodeLlama 13B |
| 30B+ | 32 GB | 20 GB | Llama 3 70B (quantized) |
| 70B+ | 64 GB+ | 40 GB+ | Llama 3 70B (full) |
* GPU with VRAM can significantly speed up local model inference. Apple Silicon Macs can use unified memory efficiently.
- ✓Gateway Port (18789) — OpenClaw gateway runs on port 18789
- ✓Outbound HTTPS (443) — For AI API calls (if using cloud AI)
- ✓Inbound (optional) — Only if hosting webhooks for certain integrations
- ✓WebSocket support — For real-time messaging (Telegram, Discord)
- ✓Stable connection — OpenClaw auto-reconnects but benefits from reliability
- ✓Config Path — ~/.openclaw/openclaw.json
- ✓Data Directory — ~/.openclaw/data/
- ✓Logs — ~/.openclaw/logs/
Mac Mini M2
Best Value16GB RAM, 256GB SSD. Perfect balance of performance, power efficiency, and silence. Can run 7B-13B models locally with great speed.
$5-10 VPS
BudgetDigitalOcean, Hetzner, or Vultr. Use cloud AI APIs. 24/7 uptime, accessible anywhere.
Gaming PC / Workstation
Power User32GB+ RAM, RTX 3080+. Run large local models (30B-70B) with GPU acceleration.